Fibroblast activation protein expression in Crohnʼs disease strictures

To the Editor: Crohn's disease (CD) is a chronic gut inflammation which, in more than one-third of cases, is complicated by intestinal fibrosis and stricture formation, frequently requiring surgical resection. While the mechanisms of inflammation in CD have been extensively investigated, knowledge of stricture pathogenesis is still limited. Intestinal strictures in CD are thought to be the result of chronic transmural inflammation combined with a dysregulated wound-healing process. Intestinal myofibroblasts are key cells in the fibrogenic process in CD, and their action is modulated by a number of cytokines, such as transforming growth factor (TGF)-β, interleukin (IL)-1β, tumor necrosis factor (TNF)-α, and IL-6.1,2 Fibroblast activation protein (FAP) is a 170-kDa type II cell surface glycoprotein belonging to the post-prolyl dipeptidyl aminopeptidase enzyme family.3 FAP is expressed on the stroma of almost 90% of human epithelial cancers and on activated fibroblasts during wound healing, but not in normal adult tissues. It is supposed to play a key role in tumor growing and wound repair through both its dipeptidyl peptidase and gelatinase activities, thus facilitating cell invasion into the extracellular matrix.4 Recently, FAP overexpression has been reported in a number of chronic inflammatory disorders characterized by a fibrotic evolution, such as idiopathic pulmonary fibrosis,5 cirrhosis,6 and rheumatoid arthritis.7 In order to assess FAP expression in the colon of fibrostenosing CD patients, we collected surgical specimens from uninflamed strictured and nonstrictured colonic areas of 15 patients with fibrostenosing CD (mean age 37.4, range 29-63). Diagnosis of CD was ascertained according to the usual clinical criteria, and strictured areas were identified by conventional or computed tomography (CT) enteroclysis. Surgical specimens from inflamed colonic areas of 15 ulcerative colitis (UC) patients undergoing colectomy were included in the study as disease controls. Resected pieces were also collected from macroscopically and microscopically unaffected colonic areas of 15 patients undergoing colectomy for colon cancer. Each patient who took part in the study was recruited after appropriate local Ethics Committee approval and informed consent was obtained in all cases. FAP expression was evaluated by immunohistochemistry on cryostat sections of full-thickness specimens and by immunoblotting on myofibroblasts that were isolated from the submucosa of colonic surgical tissues as previously described.2 Myofibroblasts were incubated with or without recombinant human TNF-α or TGF-β1 and analyzed for FAP expression by flow cytometry. The level of FAP expression on cell lines was determined as the percentage of FAP+ myofibroblasts or mean fluorescence intensity (MFI) with subtraction of values for isotype-matched controls. Data were analyzed using the paired t-test and the Mann-Whitney U-test. A level of P < 0.05 was considered statistically significant. Immunohistochemistry showed that FAP is strongly expressed in the submucosa of CD strictures (Fig. 1A, upper left picture), while no expression was observed in the submucosa of CD nonstrictured areas (Fig. 1A, upper right picture), UC submucosa (Fig. 1A, lower left picture) and normal submucosa (Fig. 1A, lower right picture). Moreover, FAP was highly expressed in the muscle layer of CD strictured areas, but not in the muscle layer of CD nonstrictured areas, UC areas, and normal colon. FAP was not expressed in the mucosa of the four groups. When we performed immunoblotting analysis on myofibroblast lysates (Fig. 1B) we observed that FAP expression, normalized for β-actin, was significantly (P < 0.001) higher in CD strictured myofibroblasts in comparison to CD nonstrictured myofibroblasts, UC myofibroblasts, and control myofibroblasts, where FAP was absent. No significant difference was found between CD nonstrictured areas, UC tissue, and normal colon. These findings were confirmed by flow cytometry (Fig. 1C), which showed that FAP expression is significantly (P < 0.001) higher in CD strictured myofibroblasts (mean 68.6 ± 12.4%) in comparison to CD nonstrictured myofibroblasts (mean 13.5 ± 4.3%), UC myofibroblasts (mean 2.7 ± 1.1%), and control myofibroblasts (mean 2.4 ± 1.4%). No significant difference was found between CD nonstrictured areas, UC tissue, and normal colon. To explore the effect of profibrogenic stimuli on CD myofibroblasts, we then incubated for 48 hours both CD strictured and nonstrictured myofibroblasts with 10 ng/mL of TNF-α or TGF-β1 (Fig. 1D). FAP expression was significantly (P < 0.001) higher in strictured (MFI: 156 ± 48) than nonstrictured (MFI: 78 ± 35) myofibroblasts. FAP expression on strictured myofibroblasts significantly (P < 0.05) increased after incubation with TNF-α (MFI: 207 ± 55) and TGF-β1 (MFI: 221 ± 53). On the contrary, both TNF-α (MFI: 102 ± 25) and TGF-β1 (MFI: 77 ± 37) had no significant effect on FAP expression by nonstrictured myofibroblasts. (A) Immunohistochemical analysis of FAP expression in the submucosa of colonic surgical specimens taken from uninflamed strictured (upper left panel) and nonstrictured (upper right panel) areas of fibrostenosing CD patients, inflamed areas of UC patients (lower left panel), and normal areas of healthy controls (HC) (lower right panel). (B) Immunoblotting for FAP on myofibroblasts isolated from the submucosa of strictured and nonstrictured areas of CD patients, inflamed areas of UC patients, and normal areas of HC. Blots were stripped and analyzed for β-actin as an internal loading control (upper panel). Each example is representative of experiments performed in 15 patients with fibrostenosing CD, 15 patients with active UC, and 15 HC. Lower panel: densitometry of Western blots normalized for β-actin. Results are means (SD). *P < 0.001 versus nonstrictured CD, UC, and HC. (C) Histograms: flow cytometric analysis of myofibroblasts isolated from uninflamed strictured and nonstrictured colonic areas of fibrostenosing CD patients, inflamed areas of UC patients, and normal areas of HC subjects. White histograms represent isotype control, while gray histograms correspond to myofibroblast FAP expression. Lower panel: FAP expression shown as percentage of FAP+ myofibroblasts. Results are means (SD). *P < 0.001 versus HC, UC, and nonstrictured CD. (D) Influence of TNF-α and TGF-β1 on FAP expression by myofibroblasts isolated from strictured and nonstrictured colonic areas of fibrostenosing CD patients. MFI, mean fluorescence intensity. Values are means (SD). *P < 0.001 versus nonstrictured CD; §P < 0.05 versus medium only. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.] In conclusion, here we show that FAP is overexpressed in CD strictures, and is upregulated on CD strictured myofibroblasts by profibrogenic stimuli. The recent availability of small-molecule FAP-inhibitors or monoclonal anti-FAP antibodies8 makes FAP a potential therapeutic target in fibrostenosing CD. This study was supported in part by the European Union framework programme 7 collaborative project grant IPODD (call: FP7-Health-2007-A).